[PATCH 1/1] mm/readahead: simplify page_cache_ra_unbounded loop counter reset

[PATCH 1/1] mm/readahead: simplify page_cache_ra_unbounded loop counter reset

[Frederick Mayle]

Minor cleanup, no behavior change intended.

`read_pages` ensures that `ractl->_nr_pages` is zero before it returns,
so the `ractl->_nr_pages` term in these expressions contributes nothing.
This seems to have been true since the statements were introduced in
commit f615bd5c4725f ("mm/readahead: Handle ractl nr_pages being
modified").

The new expression has an intuitive explanation. When filesystems
perform readahead, they increment `ractl->_index` by the number of pages
processed, so, after `read_pages` returns, `ractl->_index` points to the
first page after those already processed. `index` points to the first
page considered in the loop. So, `ractl->_index - index` is the number
of pages processed by the loop so far.

Signed-off-by: Frederick Mayle <fmayle@google.com>
---
 mm/readahead.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/mm/readahead.c b/mm/readahead.c
index 7b05082c89ea..b44f13ce5935 100644
--- a/mm/readahead.c
+++ b/mm/readahead.c
@@ -270,7 +270,7 @@ void page_cache_ra_unbounded(struct readahead_control *ractl,
 			 */
 			read_pages(ractl);
 			ractl->_index += min_nrpages;
-			i = ractl->_index + ractl->_nr_pages - index;
+			i = ractl->_index - index;
 			continue;
 		}
 
@@ -286,7 +286,7 @@ void page_cache_ra_unbounded(struct readahead_control *ractl,
 				break;
 			read_pages(ractl);
 			ractl->_index += min_nrpages;
-			i = ractl->_index + ractl->_nr_pages - index;
+			i = ractl->_index - index;
 			continue;
 		}
 		if (i == mark)

base-commit: 88ca94c8849696985b1b70ebec4b01c4b58cafdd
-- 
2.54.0.545.g6539524ca2-goog

Re: [PATCH 1/1] mm/readahead: simplify page_cache_ra_unbounded loop counter reset

[Andrew Morton]

On Thu, 30 Apr 2026 18:19:07 -0700 Frederick Mayle <fmayle@google.com> wrote:

> Minor cleanup, no behavior change intended.
> 
> `read_pages` ensures that `ractl->_nr_pages` is zero before it returns,

So it seems, but depending upon this might be a bit fragile?

It would be better to make this a more explicit/formal part of the
read_pages() contract.  kerneldocifying read_pages() would be a
suitable way.

> so the `ractl->_nr_pages` term in these expressions contributes nothing.
> This seems to have been true since the statements were introduced in
> commit f615bd5c4725f ("mm/readahead: Handle ractl nr_pages being
> modified").
> 
> The new expression has an intuitive explanation. When filesystems
> perform readahead, they increment `ractl->_index` by the number of pages
> processed, so, after `read_pages` returns, `ractl->_index` points to the
> first page after those already processed. `index` points to the first
> page considered in the loop. So, `ractl->_index - index` is the number
> of pages processed by the loop so far.
>
> ...
>
> --- a/mm/readahead.c
> +++ b/mm/readahead.c
> @@ -270,7 +270,7 @@ void page_cache_ra_unbounded(struct readahead_control *ractl,
>  			 */
>  			read_pages(ractl);
>  			ractl->_index += min_nrpages;
> -			i = ractl->_index + ractl->_nr_pages - index;
> +			i = ractl->_index - index;
>  			continue;
>  		}
>  
> @@ -286,7 +286,7 @@ void page_cache_ra_unbounded(struct readahead_control *ractl,
>  				break;
>  			read_pages(ractl);
>  			ractl->_index += min_nrpages;
> -			i = ractl->_index + ractl->_nr_pages - index;
> +			i = ractl->_index - index;
>  			continue;
>  		}
>  		if (i == mark)
> 

Re: [PATCH 1/1] mm/readahead: simplify page_cache_ra_unbounded loop counter reset

[Frederick Mayle]

On Fri, May 1, 2026 at 6:11 AM Andrew Morton <akpm@linux-foundation.org> wrote:
>
> On Thu, 30 Apr 2026 18:19:07 -0700 Frederick Mayle <fmayle@google.com> wrote:
>
> > Minor cleanup, no behavior change intended.
> >
> > `read_pages` ensures that `ractl->_nr_pages` is zero before it returns,
>
> So it seems, but depending upon this might be a bit fragile?
>
> It would be better to make this a more explicit/formal part of the
> read_pages() contract.  kerneldocifying read_pages() would be a
> suitable way.

I agree. I've sent the following patch to document the invariant
https://lore.kernel.org/r/20260508015402.735441-1-fmayle@google.com/

Re: [PATCH 1/1] mm/readahead: simplify page_cache_ra_unbounded loop counter reset

[Andrew Morton]

On Thu, 7 May 2026 19:01:27 -0700 Frederick Mayle <fmayle@google.com> wrote:

> On Fri, May 1, 2026 at 6:11 AM Andrew Morton <akpm@linux-foundation.org> wrote:
> >
> > On Thu, 30 Apr 2026 18:19:07 -0700 Frederick Mayle <fmayle@google.com> wrote:
> >
> > > Minor cleanup, no behavior change intended.
> > >
> > > `read_pages` ensures that `ractl->_nr_pages` is zero before it returns,
> >
> > So it seems, but depending upon this might be a bit fragile?
> >
> > It would be better to make this a more explicit/formal part of the
> > read_pages() contract.  kerneldocifying read_pages() would be a
> > suitable way.
> 
> I agree. I've sent the following patch to document the invariant
> https://lore.kernel.org/r/20260508015402.735441-1-fmayle@google.com/

Thanks.  I suggest you turn both into a single patch and resend when
people have returned and recovered from the conference ;)